Simple Mechanisms and Behavioral Agents: Towards a Theory of Realistic Mechanism DesignPublic Deposited
Classic mechanism design studies the implementation of optimal systems or processes in the presence of multiple selfish and fully rational agents. The theory of mechanism design has been one of the most celebrated advances in the economics discipline, especially with regards to auctions Myerson (1981),Vickrey (1961), Clarke (1971), Groves (1973). Nevertheless, practical applications of mechanism design show a significant disconnect between theory and practice. For instance, anonymous second price auction and pricing is one of the most prevalent forms of online auctions despite what the theory of Myerson (1981) suggests for asymmetric environments, e.g. eBay auctions where bidding history and demographic information provides a very specific information to the seller. On another note, most behavioral experiments indicate that the agent behavior diverges from the rational agent model, especially in dynamic environments. This thesis addresses this phenomenon by studying simple, widely prevalent auctions like anonymous pricing and developing and analyzing them through the lens of the theory of approximation. The first contribution involves the problem of selling a single to agents with independent but non-identically distributed values. We show that anonymous pricing is a constant approximation to the optimal auction. Our second contributions involves repeated sales of a non-durable good. There, economic theory suggests that in the absence of commitment power the seller generates no revenue. Our contribution was to construct an equilibrium for multiple buyers where the seller is constraint to not discriminate with constant revenue approximation. Finally, this thesis considers the incorporation of a behavioral model, particularly a model for procrastination: present-bias. We introduce the notion of variable present bias and analyze the performance loss of a procrastination agent versus a rational counterpart.